This study extends literature on recidivism after teen court to add system-level variables to demographic and sentence content as relevant covariates. Interviews with referral agents and survival analysis with proportional hazards regression supplement quantitative models that include demographic, sentencing, and case-processing variables in a database of 648 youths participating in an Illinois teen court over an 8-year period. Findings suggest that sentencing and referral-based models support an interpretation in which judgment of the offender's character is central, and post hoc analysis points to possible net-widening effects for certain teen court youths. (Contains 1 figure, 2 tables and 3 notes.)